Object Tracking Method Based on Continuous Spatiotemporal Confidence Map and Semi-supervised Extreme Learning Machine

Abstract

The invention discloses a target tracking method based on a continuous spatiotemporal confidence map and a semi-supervised extreme learning machine. The method considers that the video image frames are continuous in time, and at the same time, the position of the target to be tracked will not change abruptly. In addition, the video image Frames are also spatially continuous. Spatial continuity is reflected in the fact that there is a certain relationship between the target and the background around the target. When the appearance of the target changes greatly, this relationship can help distinguish the target to be tracked from the background area. Aiming at the problems of deformation and occlusion, the present invention fully considers the information provided by the real target, fully excavates the distribution similarity of labeled samples and unlabeled samples, improves the tracking accuracy, and proposes a method for mining labeled samples and unlabeled samples. The semi-supervised tracking method based on the extreme learning machine of distribution similarity combines the above two methods in a coupled tracking framework, and the present invention realizes a tracking with good robustness and high robustness.

Description

technical field [0001] The invention relates to a target tracking method based on a continuous space-time confidence map and a semi-supervised extreme learning machine, and belongs to the technical field of intelligent information processing and target tracking. Background technique [0002] Object tracking is an integral part of most vision systems. In specific scene applications (such as video surveillance and other fields), automatic, fast, and highly robust object tracking has attracted attention. It has broad application prospects in video surveillance, traffic detection, intelligent robots, and submarine target detection and tracking. [0003] Target tracking is an extremely important part of the field of computer vision. The moving object tracking algorithm in the video is to analyze the information of each frame of the video image in the video image sequence to be tracked, perform data mining in the video, learn the target behavior and perform a large number of Mot...

AI Technical Summary

Problems solved by technology

[0006] The present invention considers that video image frames are continuous in time, and the continuity of time is reflected in that the target to be tracked will not change greatly between adjacent frames, and the position of the target to be tracked will not change suddenly; at the same time, the video image frame is It is also continuous in space. Spatial continuity is reflected in the existence of a certain relationship between the target and the background around the target. When the appearance of the target changes greatly, this relationship can help distinguish the target to be tracked from the background area. It is proposed to use continuous space-time confidence The tracking method of graph learning overcomes the problems of poor real-time performance and robustness, lack of target spatio-temporal position information, and unobvious target features.

Images